/sys/doc/ Documentation archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Inferno

Reposted from Anssi Porttikivi <anssi.porttikivi@research.nokia.com> in
comp.lang.java. He's a subscriber to this list, though for some reason my
mail to him keeps bouncing back.

This may stir some conversation ...

From: Anssi Porttikivi <anssi.porttikivi@research.nokia.com>

Hi!

There has been discussion in comp.lang.java about the Bell 
Labs Inferno (See <http://plan9.bell-labs.com/inferno/>, 
<http://www.lucent.com/inferno/>, <news:comp.os.plan9> and 
<http://cruel.com/vanni/>) operating evironment 
(runs on different raw hardware and on 
top of MS-Windows and popular Unix versions) and it's Limbo 
language, a Java like system.

Whatever the engineering or commercial assets of Limbo 
vs. Java are, that is not the point. If it where, Java 
would win the market share battle, because Limbo as an 
interpreted or on-the-fly compiled language can't be 
so fundamentally different and Java already has really 
strong support everywhere. I see thta the real problem 
Inferno/Plan 9 targets is bigger: the state of the
computer and telecommunications industry.

Unix is an architecture which is beautiful, but 25 years 
old (an amazing achievement from the Bell people, anyway). 
MS-Windows is a collection of only moderately good ideas 
bought second hand and is as much on the way of good software 
than enabling it. (The success of MS-DOS was based on the 
fact that it was so small it was only little on the way.) 

We need a multi-purpose, modern, network centric OS. There 
are other systems which claim to be that, but they have no 
scientific muscle of Bell Labs or commercial muscle of 
Lucent to fight Sun, IBM and Microsoft operating 
systems.

The problem is not only software. The stagnation in the 
OS market hinders development in hardware. Unix enabled the 
RISC revolution, but the Wintel monopoly is stronger than 
ever and becoming more so. Inferno allows for true 
innovation in hardware, including long due multiprocessor 
architectures, the support of which goes deep and wide in 
Inferno programming tools. 

Any new interpreted system won't do: to establish a new 
standard for true hardware independency we need a small, 
efficient but higly expressive common environment, so that 
programmers don't feel tempted to bypass it. As always, 
hardest work with Inferno has probably been designing 
things out: no threading, C++ like objects 
or a microkernel - no buzzword compliance. But the same 
functionality is there implemented in better, user, 
programmer and physical reality compliant ways.

And finally, Inferno can exist in harmony with legacy 
systems: you can start using full power of it without 
giving up anything you already have.

regards
Anssi
An obsessed Inferno fan